分析了基于小波分解及局部区域能量的图像融合算法的优缺点,并针对该算法对存在局部噪声的图像以及局部噪声和局部模糊并存的图像融合效果不理想的问题,提出了改进算法。新算法利用中值滤波判断出的噪声点,小波分解后得到高频分量上得到噪声区域,对噪声区域及非噪声区域采用不同的融合规则,很好地弥补了原算法的缺陷。实验证实,改进算法对不同类型的图像具有较好的适应性和鲁棒性,有实用价值。
An image fusion method based on wavelet decomposition and energy of local area is analyzed. To overcome its disadvantage, an improved method is presented, which is more effectively on fusion of images containing local noise or both local nose and blur. By the new method, noise points are identified, noise areas in high frequency components after wavelet transformation are found, and different fusion rule is used to noise areas from non-noise areas. It is illustrated that the new algorthim is robust and available by results of experiment.